Post on 15-Oct-2020
SLUM MAPPING: MAPS AS INFRASTRUCTURES OR INFRASTRUCTURING MAPPING?
RICHARD SLIUZASGEOMUNDUS, 4-5 NOVEMBER 2016
CONTENT
Slums, slum dwellers and slum mapping requirements Infrastructure Slum mapping methods and examples Infrastucturing slum mapping?
The nature of slum dwellers and slums
Who are slums dwellers?Urban households lacking at least 1 of the following: Adequate water Adequate sanitation Sufficient living space Secure tenure Durable housing (quality
of structures & environment – hazards)
UN-HABITAT 2002
SLUMS: spatial concentration of slum dwellers -diversity of physical forms and settings
Kisumu Kenya
CairoEgypt
AhmedabadIndia
KampalaUganda
How can we define slums spatially and characterize their conditions from aerospace images?
Emphasis on slum indicator: durable housing Description of settlement extent and pattern Dwelling size and type Roof materials – quality? Orientation of buildings and road network Absence/presence of vegetation Identification of hazardous locations Natural hazards – requires specialist knowledge and data (eg
DEM, drainage, rainfall, soil, slope stability, etc) High-risk zones through mapping of traffic infrastructure and
industrial sites – needs knowledge of production processes, waste disposal and good transport routes, volumes etc.
IS IT SO STRAIGHTFORWARD?COMMON PHASES OF SLUM DEVELOPMENT
Infancy: initial land occupation and construction (bridge-headers) Consolidation: Bridge-headers expand, improve housing and
individual services; increasing subdivision, construction and tenancy
Maturity: private space maximised at expense of public space; total build out, congestion and infrastructure improvement only at expense of demolition
Formal slum development result of different process associated with filtering and physical decline of buildings and infrastructure
SLUM DEVELOPMENT PROCESS
Key aspects for mapping purposes Speed of development Level of spatial order (access as well as building layout)
Other aspects Building type Building size Density Road networks and access
Changes in the above and more
Morphological features
Slums /informal Formal /planned
Size • Small (substandard) building sizes
• Generally larger building sizes
Density • High densities (high roof coverage)
• Lack of public spaces within or in the vicinity of residential areas
• Low – moderate density areas
• Provision of public spaces within or in vicinity of residential areas
Pattern • Organic layout structure (disorderly road networkand noncompliance with planning standards)
• Regular layout pattern (orderly road network and compliance with planning standards)
The physical nature of slums: possible morphological characteristics of slum areas and planned areas
Aspect Slum settlement level City wide level
Spatial level Building – settlement Settlement-environs-city
Spatial resolution Very high (< 1m) Medium – Very high
Temporal resolution
Annual or more frequent e.g. upgrading projects
1-5 years depending on growth rates and capacity
Spectralresolution etc.
Generally optical (possiblyalso oblique or terrestrial, LIDAR)
OpticalVHR Radar (DLR)
Applications Topographic mappingRegularization of tenureSettlement upgradingVulnerability assessment
Topographic mappingUrban (slum) policy and strategic planning Monitoring (modelling)Urban growth monitoringVulnerability assessment
REQUIREMENTS
INFRASTRUCTURE
Some definitions Oxford Dictionary: A collective term for the subordinate parts of an
undertaking; substructure, foundation;
Larkin: “built networks that facilitate the flow of goods, people or ideas and allow for their exchange over space .... providing the undergirding of modern society, and they generate the ambient environment of everyday life.”
So are (slum) maps infrastructures? Or can they be conceived of as such? And is there a difference between paper maps and digital maps in this respect?
WHAT IS INFRASTRUCTURE?
Star and Ruhleder take a different view - ethnographic perspective.: “a thing becomes a tool in practice” – when it is used.
“.. infrastructure is something that emerges for people in practice, connected to activities and structures.”
So the question is not “what” but “when” and I would argue how? “As we learn to rely on electricity for work, our practices and
language change, we are “plugged in” and our daily rhythms shift. The nature of scientific and aesthetic problems shift as well.”
Bateson: “What can be studied is always a relationship or an infinite regress of relationships. Never a ‘thing.”
WHEN IS AN INFRASTRUCTURE?
Embeddedness. Transparency. Reach or scope. Learned as part of membership. Links with conventions of practice. Embodiment of standards Built on an installed base. Becomes visible upon breakdown.
KEY CHARACTERISTICS OF INFRASTRUCTURE STAR AND RUHLEDER
SOME EXAMPLES OF SLUM MAPPING
Geomundus 2016
Data sourcesPlatforms: from space to unmanned aerial vehicles, terrestrial including analogue methods.
Trends in geo-spatial technologiesIncreasing spatial, spectral and temporal resolutions
WorldView 2 - 2009 VNIR: 1.8 m
(8 bands)
PAN: 0.5 m
WorldView 3 - 2014 VNIR: 1.24 m
(8 bands)
PAN: 0.31 m
SWIR 3.7 m
CAVIS 30 m
Trends in geo-spatial technologiesUltra-high resolution imagery from UAVs
Source: Caroline Gevaert, Faculty ITC.Agatare slum, Kiglai Rwanda
Trends in geo-spatial technologiesOpen data and software
Increasing availability of open data – e.g. promoted by Group on Earth Observation – many countries now make certain types of images freely available but not VHR images
Internet platforms for access to data and maps: OpenStreetMaps, Google Earth/Maps/Map maker, India’s Buhvan server and Wikimapia etc.
Numerous open source software packages (e.g. QGIS, ILWIS, GRASS, etc.)
Advanced methods for data fusion, dense image matching, etc. to improve data extraction and quality
Slum map makers - 4 broad communities
Professional commercial data &mapping services
Academic researchers
new technologies & methods
NGO/CBOparticipation
& empowerment
Professional public data &
mapping services
How are slums being mapped?
How are slums being mapped?Research: object based data extraction
Environs
Settlement
Base ObjectBuilding Characteristics
Road Layout
Shape
Density
Neighborhood Characteristics
Location
Spatial order
Kohli, D., Sliuzas, R. V., Kerle, N., & Stein, A. (2012). An ontology of slums for image-based classification. CEUS, 36(2), 154-163.
Geographic object based image analysis
Image
Segmentation: spectral, spatial properties, scale
Classification , cleaning
Object based classification of slums in Pune India
Correctly identifiedSlums (light blue)
Confusion in Historic citycentre area
Source: Kohli et el:2014
Source: Claudio Persello
SOME PRELIMINARY RESULTS IN DAR ES SALAAM
Participatory slum mappingNGO/CBO approaches
Informal settlements in Kawempe Division Kampala
http://civitas.santafe.edu/viz/
GTZ PUMP Project Cairo
Participatory Voluntary Geo-Information (VGI) using UAV images and open source mapping tools. http://ramanihuria.org/
ACADEMIC RESEARCHIDENTIFY BUILDINGS, TERRAIN, VEGETATION, OTHER OBJECTS
32
Input data Feature extraction
Classification algorithm Map
PhD research: Caroline Gevaert. Supervisors: Prof. G. Vosselman, C. Persello, R. Sliuzas
FEATURE EXTRACTION FRAMEWORK
34
2D - Orthomosaic 2.5D - DSM 3D - Point Cloud
Feature Extraction
Feature Extraction
Feature Extraction
Classification Algorithm (SVM) using different feature sets
3D FEATURES (POINT CLOUD)SPATIAL BINNING + PLANAR SEGMENTS + POINT-BASED
37
Spatial binning
Planar segments
Point-based
• # of points• Height difference• Height std. dev.
• # of points• Avg. residual• Inclination angle• Max. height
difference to surrounding points
• Planarity, linearity, scattering, etc. of local neighborhood
FEATURE EXTRACTION - RESULTSCOMPARISON SETTLEMENTS IN RWANDA AND URUGUAY
38
RGB image 2D 2D + DSM 2D + 3D
Rw
anda
Uru
guay
Buildings Vegetation Terrain Structures Clutter
FEATURE EXTRACTION - RESULTSQUANTITATIVE EVALUATION
39
Input Feature set N Overall Accuracy (%)
5-class 10-class
Orthomosaic Radiometric 7 81.0 71.3
Radiometric + Texture 75 78.6 62.4
Radiometric + Texture over segments 82 90.8 81.8
Orthomosaic + DSM Radiometric + DSM Tophatfilters 20 86.9 77.7
Orthomosaic + Pointcloud
Radiometric + Texture and 3D over segments 103 91.5 84.0
Radiometric + Texture and 3D over segments+ Feature selection
34 91.1 82.9
APPROPRIATE CLASSIFICATION ALGORITHMSMULTIPLE KERNEL LEARNING (MKL) – SUPPORT VECTOR MACHINES
40
5.2% improvement over single-kernel SVM and 4.1% over random forests
RGB image SVM MKL-SVM Reference
Buildings High vegetation Bare surface Clutter
Walls Low vegetation Impervious surface Unlabeled
CLASSIFICATION RESULTSEXTENDED STUDY AREA (KIGALI, RWANDA)
41
REFLECTIONS ON INFRASTRUCTURINGSLUM MAPPING
Innovatieve Geo-Inwinning, Deventer, 26 mei 2011
Most slum mapping is done on a snapshot basis – time-series may or may not be made depending on context, user needs and long term commitment (organizational capacity)
The process of slum mapping is both socially and technically defined and therefore contextually bound Which areas are mapped and which not and why?
Who decides how slums are defined?
What mapping technologies and approaches are used – images, terrestrial?
Who maps, how and what?
What quality and precision requirements are used – do they meet official standards?
Who has access to the data?
How is the data used?
etc.
Geomundus 2016
INFRASTRUCTURING SLUM MAPPING
Embeddedness – slum mapping often not a structural component of mapping and related processes.
Transparency – distinction between bottom up and top-down approaches
Reach or scope – specific slums more likely than city-wide, regular updating may be unlikely.
Learned as part of membership – bottom-up more inclusive and community learning: various aspects – site, place in the city, mapping methods.
SOME IMPLICATIONS OF KEY CHARACTERISTICS OF INFRASTRUCTURE
Links with conventions of practice: community participation approaches shapes the bottom-up approach while professional practice conventional mapping but both draw on scientific developments related to new Geo-spatial technologies.
Embodiment of standards: wide use of space images, UAV images and access to GI and EO software allows bottom up approaches to meet local standards but there may be conflicting views of mapping requirements related to content and accuracy.
Built on an installed base – must adapt to existing standards, fit to other local initiatives e.g. resident saving schemes, immunization programmes, land tenure registration.
Becomes visible upon breakdown: subject to intermittent power supplies, need for regular updating to maintain usefulness.
SOME IMPLICATIONS OF KEY CHARACTERISTICS OF INFRASTRUCTURE
SOME REFERENCES
Gevaert, C., Sliuzas, R., Persello, C., & Vosselman, G. (2015). Opportunities for UAV mapping
to support unplanned settlement upgrading. In GeoTechRwanda 2015 (pp. 1–5). Kigali.
Kuffer, M., Pfeffer, K., & Sliuzas, R. (2016). Slums from Space — 15 Years of Slum Mapping
Using Remote Sensing. Remote Sensing, 8(6), 1–29. http://doi.org/10.3390/rs8060455
Larkin, B. (2013). The politics and poetics of infrastructure. Annual Review of Anthropology,
42, 327–343.
Neumann, L. J., & Star, S. L. (1996). Making Infrastructure : The Dream of a Common
Language. Proceedings of the Participatory Design Conference PDC’96, 231–240.
Star, S., & Ruhleder, K. (1996). Steps Toward an Ecology and Access of Infrastructure : for
Large Spaces Information. Information Systems Research, 7(1), 111–134.
http://doi.org/10.1287/isre.7.1.111
THANKS FOR YOUR ATTENTION
Geomundus 2016